Passive optical networks are generally made up of at least one center, the so-called central office, and a plurality of passive optical nodes. The nodes are connected to the central office(s) by optical fibers. Various network topologies can be used for this purpose, such as star network, ring network, tree structure, etc. To minimize failures, each node is connected to a central office by at least two different optical fibers. Over a main path, optical signals are transmitted to the node. When a fault, e.g. a cable break, occurs in the main path, the optical signals are transmitted to the nodes over a reserve path. When the fault has been removed, switchback to the main path is effected and the optical signals are again transmitted to the nodes over the main path.
One possibility of implementing this switchover is disclosed in EP 0973288. There, each node comprises an electric switch which is controlled by a detector and, in one preferred embodiment, by a control unit, and which is switched from the main path to the reserve path when a fault occurs in the main path. An optical coupler is provided which taps e.g. 10% of the received signal level from the main line and transfers it to the detector, which is implemented as a threshold detector or comparator, for example. The power needed by the control unit is supplied from a local power source.